Download Improved Clonal Selection Algorithm (ICLONALG)

International Journal of Current Engineering and Technology
©2015INPRESSCO®, All Rights Reserved
E-ISSN 2277 – 4106, P-ISSN 2347 – 5161
Available at http://inpressco.com/category/ijcet
Research Article
Improved Clonal Selection Algorithm (ICLONALG)
Nidhi Rai†* and Archana Singh†
†CSE
Dept., Sam Higginbottom institute of Agriculture, Technology and Sciences, Allahabad, U.P., India
Accepted 10 July 2015, Available online 14 July 2015, Vol.5, No.4 (Aug 2015)
Abstract
Natural immune system uses clonal selection algorithm to define the basic features of an immune response to an
antigenic stimulus. It establishes the idea that only those cells that recognize the antigens are selected to proliferate.
The selected cells are subjected to an affinity maturation process, which improves their affinity to the selective
antigen. In this paper, we propose a computational implementation of the clonal selection principle that explicitly
takes into account the affinity maturation of the immune response. The general algorithm, named CLONALG, is
primarily derived to solve optimization problems, emphasizing multimodal and combinatorial optimization. In this
paper there is some modification in the selection and reproduction process to maximize the optimized result.
Keywords: Artificial immune system, clonal selection theory, clonal selection algorithm, CLONALG, optimization.
1. Introduction
1 Artificial
Immune System (AIS), a new branch in
computational
intelligence
inspired
by
the
immunological principles. In the last decades, the study
of artificial immune system has been rapid increasing
interest to the large number of its possible applications
available in the field of science and engineering. The
AIS aim at using ideas from immunology in order to
develop systems capable of performing different tasks
in various area of research.
Based on the learning and evolutionary principle in
the adaptive immune response, De Castro developed
CLONal selection ALGorithm (CLONALG), which uses
individual mutation to perform greedy search and uses
random search to explore the whole solution space.
CLONALG emphasizes proportion selection and clone,
which can hold the optimal solutions by set a memory
cel. The essence of the clonal operator is producing a
variation population around the parents according to
their affinity, and then the searching area is enlarged.
In brief, the principle of theory is that the antigen (the
foreign molecule that the immune system is defending
against) selects those lymphocytes (B-Cells or white
blood cells that detect and stop antigens) with
receptors capable of reacting with a part of antigen.
Selection results in the rapid proliferation of the
selected cell to combat the invasion (clonal expansion
and production of antibodies). During this cell
duplication process coping errors occur (somatic
hypermutation) which may result in an improved
affinity of the progeny cells receptors for triggering
*Corresponding author Nidhi Rai is a PG Scholar and Archana Singh
is working as Assistant Professor
antigens. In this paper, we improved the selection
scheme and production process, which gives better
result. It gives maximized optimized result.
We do not need to make exactly the same
phenomenon, but to show that some basic immune
principles can helps as not only to better understand
the immune system itself, but also to solve complex
engineering task.
Rest of the paper describes our work to modify and
analyze clonal selection algorithm and the results we
have found during our research. In Section II we will
discuss about artificial immune system. Section III we
will be listing previous researches related to CLONALG.
In section IV we will discuss about clonal selection
theory and algorithm. In Section V will be describing
our model. In section VI, discussion about experimental
results of our model and finally Section VII will
conclude the paper providing emphasis on future
directions
2. Artificial Immune System
Artificial Immune System( AIS) are computational
systems inspired by the principles and processes of the
vertebrate immune system .The field of Artificial
immune system (AIS) is mainly concerned with the
structure and functions of the immune system to
computational system, and investigate the application
of this system towards solving computational problems
from mathematics, engineering, and information
technology. Basically an immune system has some
properties i.e. detection, diversity, learning, tolerance,
uniqueness and recognition of foreigners
2459| International Journal of Current Engineering and Technology, Vol.5, No.4 (Aug 2015)
Nidhi Rai et al
Improved Clonal Selection Algorithm (ICLONALG)
1) Detection : Identification takes place in an immune
system when the infective fragment and sensory
receptor on lymph cell surface is bonded
chemically
2) Diversity: identification in an immune system is
related to non-self bodies of the organism, thus
the immune system has number of sensory
receptor, out of which some of lymph cell will react
with the foreign organism.
3) Learning: an immune system has the capability of
detecting an eliminating the foreign organism as
soon as possible from the human body.
4) Tolerance: the particles which are mark
themselves as self bodies are contain in the
chromosomal section.
5) Uniqueness: each individual process its own
immune system, with its particular vulnerabilities
and capabilities.
6) Recognition: of foreigners: the (harmful) molecules
that are not native to the body are recognized and
eliminated by the immune system.
more of them will be readily selected for cloning and
cloned in large numbers. In addition to proliferating
and maturing into plasma cells, the immune cells can
differentiate into long-lived memory cell. Memory cells
circulate through the blood, lymph and tissues and
when exposed too second antigenic stimulus they
commence into large immune cells (lymphocyte)
capable of producing high affinity antibody specific
antigen that once stimulated the primary response.
The main role of immune system is to protect our
body from the foreign being. The immune system has
capability to distinguish between the own constituents
of our being and foreign being which can damage us.
This foreign being is known as antigen. The main role
played by the immune system is the antibodies. When
an antigen noticed in our body then those antibodies
which can distinguish the antigen will multiply by
cloning. This procedure is termed as Clonal Selection
Theory. The mechanism of clonal selection process is
shown in fig 1.
3. Clonal Selection Mechanism
3.1 Clonal Selection Theory
Artificial immune system uses the Clonal selection
theory which is a theory postulated by Burnet, Jerne,
Talmadge, used to describe the functioning of acquired
immunity, specifically a theory to define the basic
features of an immune response to an antigenic
stimulus. Clonal selection is a form of natural selection.
Main idea of clonal selection organism that only those
immune cells that recognize the antigens are selected
to proliferate, thus being selected against those that do
not.
The main features of clonal selection theory are:
1) The newly cells are replica of their parents (clone)
which are submitted to a chromosomal mutation
chemical mechanism.
2) Evacuation of newly distinguished lymph cell
carrying self-reactive sensory receptor.
3) Development and differentiation on contact of
mature cells with antigens.
When an antibody is strongly matches to an antigen,
some sub population of its bone marrow derived cells
(B lymphocytes) respond by producing anti bodies
(Ab). Each cells secrets only one kind of antibody,
which is relatively specific for the antigen. Antibody
recognizes the antigen with certain affinity (degree of
match), the B lymphocytes will be stimulated to
proliferate (divide) and eventually mature into
terminal (non-dividing) antibody secreting cells, called
plasma cell. Proliferation of the B lymphocytes is a
mitotic process whereby the cells divide themselves,
creating a set of clones identical to the parent cell. The
proliferation rate is directly proportional to the affinity
level, i.e. the higher affinity levels of B lymphocyte, the
Figure 1 Clonal Selection Principle
3.2 Clonal Selection Algorithm
Definition 1.0: A clonal selection algorithm is
primarily focused on mimicking the clonal selection
principle which is composed of the mechanism; clonal
selection, clonal expansion, and affinity maturation via
somatic hypermutation.
The clonal selection algorithm, originally called
CSA, and now renamed to CLONALG is developed on
the concept of clonal selection theory of the immune
system. In all run of algorithm, the stopping criterion is
a predefined maximum number of generations. The
goal of algorithm is to develop a memory pool of
antibodies that represents a solution to an engineering
problem. Where, an antibody represents an element of
a solution or a single solution to a problem, and an
antigens represents an element of the problem space.
2460| International Journal of Current Engineering and Technology, Vol.5, No.4 (Aug 2015)
Nidhi Rai et al
Improved Clonal Selection Algorithm (ICLONALG)
The CLONALG proposed by De Castro can be described
as follows:
1) Randomly choose an antigen and present it to
all antibodies in the repertoire, and calculate the
affinity of each antibody;
2) Select the n highest affinity antibodies to
compose a new set;
3) Clone the n selected antibodies independently and
proportionally to their antigenic affinities,
generate a repertoire set of clones;
4) Submit the repertoire set to an affinity maturation
process, maturation is inversely proportional to
the antigenic affinity. A matured antibody
population generated;
5) Re-select the highest affinity one from this set of
mature clones to be a candidate to enter the set of
memory antibodies set;
6) Finally, replace the d lowest affinity antibodies by
some random ones.
De Castro and Timmis also suggest the two key
features of clonal selection algorithms are the mutation
and cloning properties. They also suggest that selection
plays important and critical role in both the strong
selective pressure during affinity maturation, and in
the selection of long lived memory cells.
Principles 1.0: The proliferation rate of each immune
cell is proportional to its affinity with the selective
antigen (higher the relative affinity, the more progeny).
Principles 1.1: The mutation suffered by each immune
cell during reproduction is inversely proportional to
affinity of the cell receptor with the antigen (higher the
relative affinity, the lower the mutation).
Thus a general clonal selection algorithm possesses the
following mechanism:
Table 1 CLONALG parameters
1.
2.
Randomly initialize pool of antibodies
Expose the pool to antigen
a. Clonal Selection
b. Clonal Expansion
c. Clonal Hypermutation
CLONALG Description and Pseudocode
Table 2 CLONALG pseudo code listing
Parameter
P
N
n
L
Description
Antibodie’s repertoire
The
fixed
antibody
repertoire size.
The number of antibodies to
select for cloning.
Bit string length for each
antibody
Nc
D
Stop condition
Affinity
Clone
Hypermutate
Number of clone created by
each selected antibody.
Random
number
of
antibodies to insert at the
end of each generation. Best
antibodies replace the d
lowest affinity antibodies in
the repertoire.
Typically a specified number
of generation or function
evaluations.
Solution evolution.
Duplication of selected bit
string.
Modification of a bit string
where the flipping of bit(it
may be single bit or multiple
bit) is governed by an affinity
proportionate
probability
distribution.
Table 3 General algorithmic model of the clonal
selection principle
P <- rand(N,L)
While Not Stopcondition Do
ForEach p of P Do
affinity(p)
EndFor
P1 <- select(P, n)
ForEach p1 of P1 Do
C <- clone(p1)
EndFor
ForEach c of C Do
hypermutation(c)
EndFor
ForEach c of C Do
affinity(c)
Endfor
P <- insert (C, n)
Pr <-rand (d, L)
P <-replace (P. d. Pr)
EndWhile
// presentation
// clonal selection
// clonal expansion
// affinity maturation
// presentation
// greedy selection
// random replacement
4. Related Work
A short survey gives number of variants to the
CLONALG algorithm, focusing on those features that
may prove interesting or useful. Some of variants of
CLONALG are presented here. White and Garret
investigated the pattern recognition version of
CLONALG and generalized the approach for the task of
binary pattern classification renaming it Clonal
Classification (CLONCLAS). To address concerns of
algorithm
efficiency,
parameterization,
and
representation selection for continuous function
optimization. Garrett proposes an updated version of
CLONALG called Adaptive Clonal Selection (ACS).
Cutello, Narzisi, et al. proposed two modified
versions called CLONALG1 and CLONALG2 with
varying elitist strategies which were raced against the
opt-IA algorithm. Dilettoso and Selerno treated
CLONALG as a niching technique and raced it against
2461| International Journal of Current Engineering and Technology, Vol.5, No.4 (Aug 2015)
Nidhi Rai et al
Improved Clonal Selection Algorithm (ICLONALG)
traditional EC niching approaches. Wang proposed a
CSA based on CLONALG with a static clone sized
applied to power filter design observing niching like
behaviors. Cruz-Cortes, Trejo-Perez, et al. investigated
CLONALG with binary and gray encoding schemes as
well as a real-valued encoding scheme with a mutation
scheme based on Gaussian and Cauchy random
numbers. Another multi-objective application of
CLONALG was proposed by Stevens, Das, et al.
Dong, Shi, et al. proposed the Immune Memory
Clonal Selection Algorithm (IMCSA) applied to
designing stack filters for noise suppression. This
extension to CLONALG used dual-binary strings in each
antibody,
selftuning
mutation
parameter,
recombination parameters and inserted memory cells
that were developed using alternative algorithms.
CLONALG has also been hybridized with many
other optimization procedures, some examples include
the following: Zuo and Fan proposed the Chaotic
Search Immune Algorithm (CSIA) that integrated
elements of the CLONLG algorithm and was applied to
the tuning Radial-Basis Functions (RBF) in real time
controller design.
10) Merge the random population generated in step(1)
and the population generated in step(13) and
replace lower affinity cells; and repeat the step (2).
5.2 Flow Chart
5. Proposed Work
After discussing the clonal selection algorithm and
reproduction process, the improvement of CSA is
straightforward. The objective of this algorithm is to
increase the affinity value of selected cells and
maximized the optimized result. In this algorithm there
is some modification in the reproduction procedure
and selection process.
5.1 Algorithm
The algorithm overview is described as follows:
1) Randomly choose an antigen and place it to
all antibodies in the repertoire, and calculate the
affinity of each antibody;
2) Select the highest affinity antibodies to compose
a new set
3) Clone (reproduction) the
selected antibody
independently and proportionally to their
antigenic affinities, generate a temporary
population of clone;
4) Submit the population of clones to an affinity
maturation(hypermutation) process, maturation is
inversely proportional to the antigenic affinity. A
matured antibody population generated;
5) Now re-selection, the highest affinity from this set
of mature clones;
6) Clone the selected antibodies equal to the
population size and proportionally to their
antigenic affinities, generate a repertoire set of
clone;
7) Submit the repertoire set to an affinity maturation
process;
8) For i=2 to 8 (for even times) re-select(i) the
highest affinity from this set of mature clones and
clone the selected antibodies equal to the
population size, submit the repertoire set to an
affinity maturation process;
9) New population generated
Figure 2 Flow Chart of proposed algorithm
By comparing the proposed algorithm with CLONAG,
we can notice that proposed algorithm reach a diverse
set of local optima solutions as compare to CLONAG.
Essentially, their coding schemes and evaluation
functions are not different, but their evolutionary
production and selection schemes are different.
6. Results
6.1 Experimental Result
A deep analysis of the proposed algorithm has been
done on the basis of coding result. It is also
maximization problem as compare to CLONALG. Table
shows iteration vise function evaluation and give
average value and maximal value of function.
Parameters are as follows:
Number of generations: 25
Population size: 50
Hypermutation probability (pm): 0.010
function used to maximized: function f(x, y) = x.sin
(2.pi.x)-y.sin (2.pi.y+pi) +1,
Table 4 Function evaluation at each iteration
Iterations
1
2
3
4
Pm
0.0100
0.0100
0.0100
0.0100
Value
of x
Value
of y
Average
value
f(x,y)
-0.71
-0.71
-0.63
-0.63
-0.54
-0.60
-0.70
-0.70
0.80
1.02
1.11
1.16
1.602
1.936
2.039
2.039
2462| International Journal of Current Engineering and Technology, Vol.5, No.4 (Aug 2015)
Nidhi Rai et al
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
Improved Clonal Selection Algorithm (ICLONALG)
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0100
0.0800
0.0800
-0.63
-0.63
-0.63
-0.63
-0.69
-0.69
-0.69
-0.69
-0.64
-0.64
-0.64
-0.64
-0.64
-0.63
-0.63
-0.63
-0.63
-0.63
-0.63
-0.63
-0.63
-0.70
-0.70
-0.70
-0.70
-0.60
-0.60
-0.60
-0.70
0.65
0.65
0.65
0.65
0.65
-0.62
-0.62
-0.62
-0.62
-0.62
-0.62
-0.62
-0.62
1.23
1.23
1.33
1.51
1.56
1.57
1.51
1.61
1.66
1.70
1.74
1.72
1.76
1.79
1.82
1.84
1.78
1.85
1.93
1.89
1.95
2.039
2.039
2.039
2.039
2.049
2.049
2.049
2.064
2.240
2.240
2.241
2.241
2.241
2.250
2.250
2.250
2.250
2.250
2.250
2.250
2.250
Figure 4 Optimized population after cell
generation
The affinity measure related to the evaluation of the
function f(x,y) after decoding the value of x and y,
figure 4 presents the optimized population after
defined generation. The solutions cover most of the
peaks, including the global optimum.
Maximum found [x, y, f(x,y)]: [-0.63, -0.62, 2.25]
Table 5 Comparison Table
6.2 Multi-Modal Optimization
Avg,f(x,y)
The individuals are reproduces by clonal selection
algorithm that has higher affinity and selects their
improved maturated progenies. This scheme suggest
that the algorithm perform the greedy search, where
single members will be locally optimized (exploitation
of the surrounding space), and the newer yield a
broader exploration of the search space. Because of
this characteristic of CSA, it is suitable for solving
modal optimization task, and as illustration, consider
the case of maximizing the function f(x, y) = x.sin
(2.pi.x)-y.sin (2.pi.y+pi) +1, shown in figure 3. This
function is composed of many local optima and single
global optimum.
Algo
Avg
f(x,y)
CLONALG
1.92
2.02
ICLONALG
1.95
2.25
f(x,y) x Mean
2.6
2.4
2.2
2
f(x)
1.8
1.6
1.4
1.2
1
0.8
0
5
10
15
Generations
20
25
30
Figure 5 Best and average fitness function
determined by CSA (simple evaluation function)
Conclusions
Figure 3 Function to be maximized by the CSA
In this paper, we proposed a general-purpose
algorithm inspired in clonal selection principle. There
is modification in the reproduction process which
improves the performance and results also. The
algorithm is capable to solve complex problems, like
multi-modal optimization. The algorithm introduced
constitutes a version of the clonal selection principle.
2463| International Journal of Current Engineering and Technology, Vol.5, No.4 (Aug 2015)
Nidhi Rai et al
Improved Clonal Selection Algorithm (ICLONALG)
By comparing the proposed algorithm with CLONAG,
we can notice that proposed algorithm reach a diverse
set of local optima solutions as compare to CLONAG.
Essentially, their coding schemes and evaluation
functions are not different, but their evolutionary
production and selection processes are different. We
do not advocate this algorithm performs better than
CLONALG, but it gives better result in maximization
problems. Instead, we demonstrate that the proposed
algorithm is derived from the CLONALG, which
performs learning and multimodal search, and
presents a fine tractability in terms of computational
cost.
References
Cziko, G. (1995), The Immune System: Selection by the
Enemy, In Without Miracles, G. Cziko, A Bradford Book, The
MIT Press, 39-48.
Dasgupta, D., (1999), Artificial Immune Systems and Their
Applications, Ed., Springer-Verlag. F.M.Burnet, The Clonal
Selection Theory of Acquired Immunity, Australian Journal
of Science, Cambridge University Press, 1959,pp.67-69.
J. H. Holland, Adaptation in Natural and Artificial
Systems, MIT Press,5th ed. 1998.
De Castro,L.N and Von Zuben,F.J (2000), The Clonal
Selection
Algorithm with Engineering
Application,
Proc.of GECCO’00,Workshop on Artificial Immune Systes
and their Application ,36-37.
Hofmeyr S. A. and Forrest, S. (1999)Immunity by Design: An
Artificial Immune System. Proc. of GECCO’99, (1999) 12891296
Leandro N. de Castro and Fernando J. Von Zuben (Jun, 2002),
Learning and optimization using the clonal selection
principle IEEE Transactions on Evolutionary Computation,
vol. 6, pp. 239-251.
Aickelin, U., Greensmith, J., and Twycross,J (2004)
Immune System Approaches to Intrusion Detection—a
Review. In the Proceeding of the Third International
Conference on Artificial Immune Systems (ICARIS–04),
pp.316–329.
Leandro N. de Castro and Fernando J. Von Zuben (2000), The
Clonal Selection Algorithm with Engineering Application,
Proceedings of the Genetic and Evolutionary Computation
Conference (GECCO '00). Workshop on Artificial Immune
Systems and Their Applications, Las Vegas, Nevada, USA,
pp. 36-37.
Leandro N. de Castro and Jon Timmis. Artificial Immune
Systems: A new computational intelligence approach,
Great Britain: Springer-Verlag.
Frank Macfarlane Burnet (1959), The clonal selection theory
of acquired immunity, Nashville, Tennessee, U.S.A.:
Vanderbilt University Press.
J. Lederberg (1988), Ontogeny of the clonal selection theory
of antibody formation. Reflections on Darwin and Ehrlich
Annals of the New York Academy of Sciences, vol. 546, pp.
175-182.
2464| International Journal of Current Engineering and Technology, Vol.5, No.4 (Aug 2015)